This directory contains a program (frequencies.py) for counting the
frequencies of non-alphanumeric characters in all the files under a
given directory tree, and some results in the "results/" directory.
This will help show what non-standard keys need to be most convenient
in a proposed "programmer's" Maltron keyboard, to be based on the L89P
layout.  To see a summary of the frequency results so far, do this:

   $ cd results/
   $ ./totals.py *.out

Answers to questions you might be about to ask:

Q: Why count only the non-alphanumerics?

A: Because we already know that the alphanumerics are the most common
   things in source code, just as they are in prose.  Well, we know
   that the alphas are most common, and the numerics aren't going to
   move from their current spots in the L89P layout anyway, so there's
   no point fussing with them.

Q: Why use only single-character counts, not digraphs and trigraphs?

A: Using digraph and trigraph data makes sense when coming up with the
   overall layout, but that's already been done here.  We're just
   tweaking the positions of a few keys at this point, and there's not
   much room left to work with. Trying to use di/trigraph data would
   only make the process more complicated than it needs to be.

Q: What about editor motion commands and other key usages not
   reflected in file contents?

A: We just have to go on knowledge of typical interfaces for that.  I
   suppose we could ask a lot of people who use different editing
   environments to tell their editors to save keystroke transcripts,
   but... there's such a thing as going overboard.